1
0
mirror of https://github.com/ytdl-org/youtube-dl.git synced 2024-11-06 01:22:35 +00:00

Move Photobucket into its own file

This commit is contained in:
Philipp Hagemeister 2013-06-23 20:12:18 +02:00
parent 219b8130df
commit 97d6faaced
2 changed files with 67 additions and 56 deletions

View File

@ -27,68 +27,13 @@ from .extractor.common import InfoExtractor, SearchInfoExtractor
from .extractor.dailymotion import DailymotionIE from .extractor.dailymotion import DailymotionIE
from .extractor.metacafe import MetacafeIE from .extractor.metacafe import MetacafeIE
from .extractor.statigram import StatigramIE from .extractor.statigram import StatigramIE
from .extractor.photobucket import PhotobucketIE
from .extractor.youtube import YoutubeIE, YoutubePlaylistIE, YoutubeUserIE, YoutubeChannelIE from .extractor.youtube import YoutubeIE, YoutubePlaylistIE, YoutubeUserIE, YoutubeChannelIE
class PhotobucketIE(InfoExtractor):
"""Information extractor for photobucket.com."""
# TODO: the original _VALID_URL was:
# r'(?:http://)?(?:[a-z0-9]+\.)?photobucket\.com/.*[\?\&]current=(.*\.flv)'
# Check if it's necessary to keep the old extracion process
_VALID_URL = r'(?:http://)?(?:[a-z0-9]+\.)?photobucket\.com/.*(([\?\&]current=)|_)(?P<id>.*)\.(?P<ext>(flv)|(mp4))'
IE_NAME = u'photobucket'
def _real_extract(self, url):
# Extract id from URL
mobj = re.match(self._VALID_URL, url)
if mobj is None:
raise ExtractorError(u'Invalid URL: %s' % url)
video_id = mobj.group('id')
video_extension = mobj.group('ext')
# Retrieve video webpage to extract further information
webpage = self._download_webpage(url, video_id)
# Extract URL, uploader, and title from webpage
self.report_extraction(video_id)
# We try first by looking the javascript code:
mobj = re.search(r'Pb\.Data\.Shared\.put\(Pb\.Data\.Shared\.MEDIA, (?P<json>.*?)\);', webpage)
if mobj is not None:
info = json.loads(mobj.group('json'))
return [{
'id': video_id,
'url': info[u'downloadUrl'],
'uploader': info[u'username'],
'upload_date': datetime.date.fromtimestamp(info[u'creationDate']).strftime('%Y%m%d'),
'title': info[u'title'],
'ext': video_extension,
'thumbnail': info[u'thumbUrl'],
}]
# We try looking in other parts of the webpage
video_url = self._search_regex(r'<link rel="video_src" href=".*\?file=([^"]+)" />',
webpage, u'video URL')
mobj = re.search(r'<title>(.*) video by (.*) - Photobucket</title>', webpage)
if mobj is None:
raise ExtractorError(u'Unable to extract title')
video_title = mobj.group(1).decode('utf-8')
video_uploader = mobj.group(2).decode('utf-8')
return [{
'id': video_id.decode('utf-8'),
'url': video_url.decode('utf-8'),
'uploader': video_uploader,
'upload_date': None,
'title': video_title,
'ext': video_extension.decode('utf-8'),
}]
class YahooIE(InfoExtractor): class YahooIE(InfoExtractor):

View File

@ -0,0 +1,66 @@
import datetime
import json
import re
from .common import InfoExtractor
from ..utils import (
ExtractorError,
)
class PhotobucketIE(InfoExtractor):
"""Information extractor for photobucket.com."""
# TODO: the original _VALID_URL was:
# r'(?:http://)?(?:[a-z0-9]+\.)?photobucket\.com/.*[\?\&]current=(.*\.flv)'
# Check if it's necessary to keep the old extracion process
_VALID_URL = r'(?:http://)?(?:[a-z0-9]+\.)?photobucket\.com/.*(([\?\&]current=)|_)(?P<id>.*)\.(?P<ext>(flv)|(mp4))'
IE_NAME = u'photobucket'
def _real_extract(self, url):
# Extract id from URL
mobj = re.match(self._VALID_URL, url)
if mobj is None:
raise ExtractorError(u'Invalid URL: %s' % url)
video_id = mobj.group('id')
video_extension = mobj.group('ext')
# Retrieve video webpage to extract further information
webpage = self._download_webpage(url, video_id)
# Extract URL, uploader, and title from webpage
self.report_extraction(video_id)
# We try first by looking the javascript code:
mobj = re.search(r'Pb\.Data\.Shared\.put\(Pb\.Data\.Shared\.MEDIA, (?P<json>.*?)\);', webpage)
if mobj is not None:
info = json.loads(mobj.group('json'))
return [{
'id': video_id,
'url': info[u'downloadUrl'],
'uploader': info[u'username'],
'upload_date': datetime.date.fromtimestamp(info[u'creationDate']).strftime('%Y%m%d'),
'title': info[u'title'],
'ext': video_extension,
'thumbnail': info[u'thumbUrl'],
}]
# We try looking in other parts of the webpage
video_url = self._search_regex(r'<link rel="video_src" href=".*\?file=([^"]+)" />',
webpage, u'video URL')
mobj = re.search(r'<title>(.*) video by (.*) - Photobucket</title>', webpage)
if mobj is None:
raise ExtractorError(u'Unable to extract title')
video_title = mobj.group(1).decode('utf-8')
video_uploader = mobj.group(2).decode('utf-8')
return [{
'id': video_id.decode('utf-8'),
'url': video_url.decode('utf-8'),
'uploader': video_uploader,
'upload_date': None,
'title': video_title,
'ext': video_extension.decode('utf-8'),
}]